I have two questions on how to apply negative feedback to an RCA input self-inverting push pull tube amplifier:
If you tap off, say, the 8-ohm secondary to the cathode of the input tube, do you have to run a parallel feedback circuit from the other taps if you want to use them?
If you want to add a second input tube to have a balanced XLR input, how do you apply feedback to the inverting tube without it seeing an out of phase feedback signal?
If you tap off, say, the 8-ohm secondary to the cathode of the input tube, do you have to run a parallel feedback circuit from the other taps if you want to use them?
If you want to add a second input tube to have a balanced XLR input, how do you apply feedback to the inverting tube without it seeing an out of phase feedback signal?
One feedback is enough even if you don't use the 8 ohm tap .
Output phase can be inverted if you swap the transformer primary ( reverse the tubes plate connections ) .
Output phase can be inverted if you swap the transformer primary ( reverse the tubes plate connections ) .
Last edited:
If the phase is the same at the cathodes of both tubes it could be done . But as you drew the schematic seems to be out of phase .
Last edited:
Balanced power output stage is rather complicated, even with solid state devices.
There would be 3 (not 2) negative feedback paths. One for positive output, one for negative output, the 3rd for keeping the middle point of 2 outputs to 0volts.
There would be 3 (not 2) negative feedback paths. One for positive output, one for negative output, the 3rd for keeping the middle point of 2 outputs to 0volts.
See this schematic. The 4 Ohm tap is grounded to provide equal and opposite output voltages.
It also has partial cathode coupling, but that is not necessary.
https://audio-circuit.dk/downloads/audioresearch/AudioResearch-D76-pwr-sch.pdf
Last edited:
You won’t get the same phase at both cathodes with a differential input. The whole point is that they are opposite phase. Best way to apply feedback to that would be with a center tapped tertiary winding on the transformer, or perhaps a second transformer which provides both phases just for feedback. You might consider local feedback on each stage instead (or output stage only) and skip the global.
In the second pic, the feedback is not negative but positive (opposite phase). It's an oscillator.
BUT if you feed that feedback to the bottom output tube cathode, which has the opposite phase from the top one, you have restored the feedback to negative and it should work.
Note that you now have the whole input part outside the feedback loop so performance will be worse.
Jan
There are advantages (and disadvantages) to taking feedback from the primary of the output transformer. Classic designs never did this, so until recently DIYers didn't either, but times are changing.
All good fortune,
Chris
All good fortune,
Chris
Menno Vanderveen showed that if you drive the output transformer from a very low impedance point, taking the feedback from the primary doesn't cause more distortion, but does make the whole thing much more stable compared to taking the output from the secondary.
https://mennovanderveen.nl/cms/imag.../Vanderveen-Trans-SE10-Secrets-AX-2021-01.pdf
Jan
https://mennovanderveen.nl/cms/imag.../Vanderveen-Trans-SE10-Secrets-AX-2021-01.pdf
Jan
I think he named it after the Transconductance-Transadmittance design of the early semicon era. Didn't stick for semicon amps but quite possibly the next phase for DIY valve amps. The "Schade" designs getting popular now are on the slippery slope.
All good fortune,
Chris
All good fortune,
Chris
Slippery in the sense of a natural gravity-fed progression from casually adding a resistor from anode to grid, to deliberately making a driving stage optimized for current output linearity (Not! necessarily a triode).
edit: and a summing junction
Thanks, as always,
Chris
edit: and a summing junction
Thanks, as always,
Chris
Exactly. Driver stage design becomes less straightforward. If it was easy, everybody would do it.
I'm getting excellent results in SE amp experiments taking voltage feedback from the primary and series-applying it at the driver cathode. I use a CCS-loaded high-gm pentode for the driver.
Distortion goes way down since there is a lot of feedback, but it stays low-order and 2nd-dominant.
I haven't found a way to get a really linear V-to-I conversion to parallel-apply the feedback at the output tube grid. At least not nearly as well as the approach above.
Besides, most output tubes don't have enough gm to get distortion really low even with 100% feedback in the output stage. If you involve the driver in the feedback system, the driver can multiply up the overall gm of the driver/output tube system. It's really the way to go if you want to push the envelope.
Distortion goes way down since there is a lot of feedback, but it stays low-order and 2nd-dominant.
I haven't found a way to get a really linear V-to-I conversion to parallel-apply the feedback at the output tube grid. At least not nearly as well as the approach above.
Besides, most output tubes don't have enough gm to get distortion really low even with 100% feedback in the output stage. If you involve the driver in the feedback system, the driver can multiply up the overall gm of the driver/output tube system. It's really the way to go if you want to push the envelope.
Applying feedback to a single ended input stage is relatively straightforward. The principle of 2nd plate to 1st cathode doesn’t really matter if the feedback is taken before or after the transformer. The phase relationship is the same. Finding a way to do it for the 2nd case in post #3 (with no phase splitter) requires out of box thinking.
Well, I was commenting on the subsequent comments from Chris and Jan, not really to the OP question. I've done ~94% feedback on a 6L6 variant with an op-amp/fet based V-to-I converter (near perfect conversion) and still ended up with over 1% distortion at 1W. A 6L6 just has too little linearity/too little gm to clean itself up enough with single stage feedback. More feedback is needed.
Van der Veen's paper is interesting. He seems to have managed to make a system work quite well, but I can tell you from experience that the aim of a V-to-I converter driver should not be good V-to-I conversion, it should be complementary V-to-I conversion to the nonlinearity of the output stage. But like I said in my other post, involving two stages in the feedback is much better. My breadboarded amp has less than half the distortion as Van der Veen's at 1W and is 2nd harmonic dominant. Whatever is happening in his V-to-I converter is cancelling almost all of the 2nd harmonic distortion. Good for him for doing that cancellation well, but the amp has lost the "SE signature" if one cares for such things.
As to the OP, the amps I have seen derive complementary feedback signals have a 4-8-16 Ohm secondary with taps so that the 4 Ohm tap can be grounded to generate two opposite phases from the secondary, like rayma points out above. Choose which phase goes where and you have a lot of feedback options.
With a single secondary transformer, you need to specifically ask for a center tap at manufacture. I have done so in the past. I've got a pair of push-pull Edcor transformers in the garage with an 8-Ohm secondary with a center tap.
It seems that the transformer is drawn with multiple taps. If one is a center tap, why not ground that and derive two opposite phases for feedback? Seems pretty straightforward. Am I missing something? I'm having trouble understanding what is meant in some of the diagrams in post 3 since things are shown going directly to GND that probably shouldn't be, otherwise we are just going to completely short the output. Is there a schematic for the actual amp in question so that I can understand what we are getting at?
Van der Veen's paper is interesting. He seems to have managed to make a system work quite well, but I can tell you from experience that the aim of a V-to-I converter driver should not be good V-to-I conversion, it should be complementary V-to-I conversion to the nonlinearity of the output stage. But like I said in my other post, involving two stages in the feedback is much better. My breadboarded amp has less than half the distortion as Van der Veen's at 1W and is 2nd harmonic dominant. Whatever is happening in his V-to-I converter is cancelling almost all of the 2nd harmonic distortion. Good for him for doing that cancellation well, but the amp has lost the "SE signature" if one cares for such things.
As to the OP, the amps I have seen derive complementary feedback signals have a 4-8-16 Ohm secondary with taps so that the 4 Ohm tap can be grounded to generate two opposite phases from the secondary, like rayma points out above. Choose which phase goes where and you have a lot of feedback options.
With a single secondary transformer, you need to specifically ask for a center tap at manufacture. I have done so in the past. I've got a pair of push-pull Edcor transformers in the garage with an 8-Ohm secondary with a center tap.
It seems that the transformer is drawn with multiple taps. If one is a center tap, why not ground that and derive two opposite phases for feedback? Seems pretty straightforward. Am I missing something? I'm having trouble understanding what is meant in some of the diagrams in post 3 since things are shown going directly to GND that probably shouldn't be, otherwise we are just going to completely short the output. Is there a schematic for the actual amp in question so that I can understand what we are getting at?
Can't really agree that phase of feedback signal is the same when taken from primary or secondary. Although the classic lumped-sum model of transformers is pretty shaky, there's got to be *some* amount of (modeled) leakage inductance which appears in series with (modeled, lump-sum) shunt capacitance.
However over-simplified the classic model may be, there's two poles of low-pass between the output valve's anode and the "magic idealized internal transformer", and therefor the secondary, somewhere up there. Not the end of civilization as we know it (that was a year ago) but just another tradeoff.
I gotta admire someone campaigning for feedback over two stages. It perfectly suits modern DIY work, by both its flexibility and its demand that the designer think about internal interactions. Worth the effort both ways.
All the best fortune,
Chris
However over-simplified the classic model may be, there's two poles of low-pass between the output valve's anode and the "magic idealized internal transformer", and therefor the secondary, somewhere up there. Not the end of civilization as we know it (that was a year ago) but just another tradeoff.
I gotta admire someone campaigning for feedback over two stages. It perfectly suits modern DIY work, by both its flexibility and its demand that the designer think about internal interactions. Worth the effort both ways.
All the best fortune,
Chris
Last edited:
Oh, I wasn’t counting the frequency dependent phase *shift*, just the number of inversions., which needs to match or nothing works. You can probably apply MORE dB of NFB before running into phase margin issues by tapping pre transformer than post. At least one less pole, which of course won’t be corrected.